DIY Raspberry Pi Spectrometer Guide

Backstory: I've repeatedly encountered deep confusion about color, even among first-tier physical-sciences graduate students. Yet color is widely taught K-2. Apparently without great success. So what might a rewrite, a modern learning progression for color, look like? Perhaps one based on spectra, a modern colorspace, and building on current understanding of color perception? Tablets are used in K - "find and take a picture of a circle". So how about using them for color? There's middle-school work with color "arithmetic" (an binary triple with addition(light) and subtraction(filter)). And phone spectrographs are a thing. Thermal IR inspection cameras suggest having a context image aids understandability, and phones now have multiple cameras, so might one do a more accessible sample-with-context spectroscope app? With the light path folded flat, not sticking out? And a high dynamic range to permit sampling objects under ambient illumination? Might one craft a spectra-based introduction to color? For K?

Beyond 1000 nm, silicon becomes transparent and ceases to work as a detector, so those longer wavelengths need a detector made from another material, notably indium gallium arsenide (InGaAs) which in one form can get all the way out to around 2700 nm. Anything that gets you away from silicon chip fab also gets you away from the fab-ulous economics of silicon. InGaAs sensors are super damn expensive.

Beyond 2700, thermal imaging cameras and the like use even more exotic sensor materials.

An alternative for those longer wavelengths is a monochromator (e.g., rotating diffraction grating detecting one wavelength at a time) and a single element detector which is cheaper than an array. If course your subject has to be sitting still for the duration of your measurement.